arXiv:2607.10744v5 Announce Type: replace Abstract: Benefiting from the powerful priors embedded in large-scale pre-training data and the emerging commonsense reasoning ability, large language models...
arXiv:2410.14606v3 Announce Type: replace Abstract: Learning from a stream of experience as it arrives, also known as streaming learning, is a core part of natural learning. However, reliable streami...
arXiv:2609.36219v1 Announce Type: new Abstract: Perspective taking is a fundamental component of spatial intelligence, requiring models interpret spatial relations from a specified viewpoint, such as...
LoopVLA introduces a recurrent Vision‑Language‑Action architecture that learns to refine multimodal representations, predict actions, and estimate when further refinement is unnecessary. By iteratively applying a shared Transformer block and producing a sufficiency score at each step, it decouples refinement from fixed layer indices and aligns confidence scores with action quality through a self‑supervised objective. Experiments on LIBERO, LIBERO‑Plus, and VLA‑Arena demonstrate that LoopVLA reduces model parameters by 45% and boosts inference throughput up to 1.7× while matching or surpassing strong baselines in task success.
arXiv:2604. 16557v2 Announce Type: replace Abstract: Current post-training methodologies for adapting Large Vision-Language Models (LVLMs) generally fall into two paradigms: Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL).
AWM‑VLA introduces a unified framework that embeds aligned world modeling directly into a diffusion‑transformer vision‑language‑action policy. By adding learnable future tokens aligned with vision‑language embeddings of future observations, the policy can anticipate long‑term consequences while generating actions. The method extends this with an object‑centric alignment objective and a principled weighting scheme, achieving up to 21% higher success rates on RoboCasa and humanoid tabletop benchmarks and producing object‑centric rationales preferred by human raters in 83% of cases.
CondenseVLA introduces a learnable history‑condensation module that distills short‑term visual context into a fixed set of slot tokens via learnable queries and cross‑attention, reducing redundancy and improving action decoding. The approach achieves notable gains on several robot manipulation benchmarks, raising Simpler‑WidowX success from 58.7% to 69.8%, reaching 97.2% on LIBERO, 52.8% on real‑world tasks, and 21.6% on memory‑dependent RMBench tasks.
arXiv:2506. 01274v2 Announce Type: replace-cross Abstract: Recent progress in Large Multi-modal Models (LMMs) has enabled effective vision-language reasoning, yet the ability to video understanding remains constrained by suboptimal frame selection strategies, albeit with the rapid development of video-specialized LMMs.
Falcon Perception-HD applies reinforcement learning (GRPO) to autoregressive perception models, aligning them directly with precision and recall metrics rather than relying on maximum‑likelihood fine‑tuning. The RL framework introduces reward design for set‑structured outputs and multi‑head sampling control, enabling state‑of‑the‑art performance in very dense scenes (up to 500 objects) and eliminating common issues such as mask repetitions, NMS, and coordinate deduplication. Hybrid self‑annotation pipelines tailored for difficult referring expressions and dense scenes further boost RL training, with improvements observed across all difficulty levels on PBench and SACO‑Gold, and the model preserves object existence knowledge without negative samples.
arXiv:2609.15851v1 Announce Type: new Abstract: Language models can learn from experience, but raw solution trajectories are often too long and noisy to provide effective guidance. In this work, we p...
arXiv:2605.11151v3 Announce Type: replace Abstract: Offline-to-online reinforcement learning (RL) improves sample efficiency by leveraging pre-collected datasets prior to online interaction. A key ch...
DVLA-RL++ introduces Dual-Level Vision‑Language Alignment with Reinforcement Learning Gating for few‑shot learning. It adds complementary semantic purification (CSP) to separate intrinsic from nuisance textual descriptions and counterfactual reinforcement‑learning gating (CRG) to learn layer‑wise fusion strengths. Experiments on standard, fine‑grained, and cross‑domain benchmarks demonstrate state‑of‑the‑art accuracy, improving on DVLA‑RL by an average of 1.4%.